INDEX

Explanations

terms related to expertise, investigation, and specialized knowledge

oai_token-act-pair · gpt-3.5-turbo

keywords associated with research and expertise

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inventoryQuantity

-0.69

ebus

-0.66

ibur

-0.61

ASA

-0.59

 Worker

-0.58

 Tactics

-0.58

 Quote

-0.56

PDATE

-0.56

uncture

-0.56

 Rampage

-0.55

POSITIVE LOGITS

 comprising

0.97

 culminating

0.97

 vying

0.93

 spanning

0.92

 converge

0.92

 conver

0.88

 including

0.86

 consisting

0.86

 collectively

0.85

 ranging

0.82

Activations Density 0.431%

terms related to expertise, investigation, and specialized knowledge

keywords associated with research and expertise

No Comments

No Known Activations